In, for example, a GPU (Graphical Processing Unit), a technique called a UMA (Unified Memory Architecture) for sharing one memory between a plurality of arithmetic processors is known.